Survey on Cross-Age Face Comparison

Cross-Age Face Comparison has become a popular research topic in computer vision and biometric because of its ability to deal with facial changes that occur during the aging process. This problem has many applications including finding a missing person, identifying criminals, government document verification, image matching, image retrieval, etc. Traditionally, different models such as Hidden Factor Analysis, Maximum Entropy Feature Descriptor, and Cross-Age Reference Coding were used by researcher to solve the problem. These models need feature extraction using Local Binary Pattern and Histogram of Oriented Gradient. On the other hand deep learning based techniques do not require feature extraction and are being widely used for facial analysis. This paper focuses on the broad literature survey of state-of-the-art researches attempting Cross-Age Face Recognition/Retrieval and Cross-Age Face verification. Major contributions of the paper include (1) analysis of various Face comparison challenges with their solutions given by researchers and (2) analysis of various traditional/deep learning based models used to solve the Age invariant problem.